Autism spectrum disorder (ASD) is a heterogeneous neurodevelopmental condition characterized by persistent difficulties in social communication together with restricted, repetitive patterns of behaviour and sensory-processing differences. Growing evidence suggests that ASD is shaped by complex interactions among genetic susceptibility, epigenetic regulation, immune signalling, maternal and early-life exposures and gut microbiome-related pathways. However, many of these associations remain biologically plausible rather than definitively causal, particularly when findings from experimental models are considered alongside human clinical data. This narrative review examines recent advances across these interconnected domains, with particular emphasis on maternal immune activation, prenatal nutrition, gut microbial imbalance, epigenetic and molecular mechanisms, emerging therapeutic directions and developing biomarker platforms. We also discuss current diagnostic limitations and evaluate the potential of salivary microRNAs, perinatal metabolic and epigenetic markers, oxidative stress-related measures and microbiome-based profiles as early and biologically informative indicators of ASD risk. Special attention is given to the need for biologically informed stratification, although current subgrouping frameworks remain preliminary and not yet sufficiently validated for routine clinical use. Likewise, candidate biomarkers remain investigational and require stronger evidence for reproducibility, external validation, longitudinal performance and clinically meaningful sensitivity and specificity before they can be considered for screening or precision-guided care. Emerging therapeutic strategies targeting immune, epigenetic and microbiome-related pathways are also reviewed, but most remain preclinical or early-stage and face substantial translational barriers. The convergence of epigenomics, microbiome research and early diagnostic science may help advance a more personalized medicine framework for ASD, provided that future studies improve cross-cohort reproducibility, clarify brain relevance of peripheral signals and develop practical multiomics models that can support clinically meaningful integration.
Sehrawy et al. (Fri,) studied this question.